M
ASTER
F
INANCE
M
ASTER
’
S
F
INAL
W
ORK
D
ISSERTATION
D
OES
CEO
TURNOVER INFLUENCE DIVIDEND POLICY
?
P
EDRO
J
OSÉ
I
NÁCIO DA
C
OSTA
S
ANTOS
M
ASTER
F
INANCE
M
ASTER
’
S
F
INAL
W
ORK
D
ISSERTATION
D
OES
CEO
TURNOVER INFLUENCE DIVIDEND POLICY
?
P
EDRO
J
OSÉ
I
NÁCIO DA
C
OSTA
S
ANTOS
S
UPERVISION
:
P
ROFESSOR
D
OUTOR
V
ICTOR
B
ARROS
i
Abstract
In this research, we aim to assess whether CEO turnover influences firms’ dividend policy.
This work is motivated by the extensive conceptualisation and empirical research that CEO turnover and dividend policy have been subject to throughout the years. However, to the best of our knowledge, there is no empirical literature that links CEO turnover and dividend policy, so far. Therefore, with this study we intend to clarify whether and how CEO turnover influences firms’ dividend policy and, consequently, contribute to an unexplored topic.
The data used in this study contains 394 firms listed in the S&P 500 Index with a sample period between 2004 and 2017. The empirical evidence suggests that CEO turnover increases firms’ dividend yield by 0.2%. Moreover, CEO turnover that occurs during 2008 and 2012 has a positive effect on the dividend yield of 0.5%, although it leads to a decrease in the dividends paid by firms. During the financial crisis stock prices are more volatile, therefore, when a firm announces a CEO turnover, the market reacts less smoothly and may lead to even lower stock prices, increasing, even more, the dividend yields. Also, during the same period (2008-2012) firms have more incentives to retain their earnings. Therefore, CEOs decide to pay fewer dividends. Evidence also indicates that CEO turnover has a positive effect on dividend per share and dividend yield after the financial crisis.
Thus, this work contributes to practice since evidences, for the first time, that CEO turnover has a significant impact on firms’ dividend policy, contributing to the existing literature of both CEO turnover and dividend policy.
JEL Classification: G32; G35; G40
Keywords: CEO turnover, Dividend Policy, Dividend Yield, Dividend Per Share, Dividend
ii
Resumo
Neste estudo pretendemos estudar se as mudanças de CEO influenciam a política de dividendos das empresas.
Este trabalho é motivado pela extensiva conceptualização e análise empírica de que as mudanças de CEO e a política de dividendos têm sido alvo ao longo dos anos. No entanto, a nosso conhecimento, não existe até agora literatura empírica que relacione mudanças de CEO com política de dividendos. Por isso, com este estudo pretendemos clarificar como é que as mudanças de CEO influenciam a política de dividendos das empresas e, consequentemente, contribuir para um tópico que ainda não foi estudado.
Os dados usados neste estudo contêm 394 empresas cotadas no S&P 500 Index com um período de amostra entre 2004 e 2017. Os resultados da análise feita sugerem que mudanças de CEO aumentam o rendimento dos dividendos das empresas em 0.2%. Além disso, mudanças de CEO que ocorrem entre 2008 e 2012 têm um efeito positivo no rendimento dos dividendos de 0.5% e levam a uma diminuição dos dividendos pagos pelas empresas. Durante esta crise financeira, o preço por ação é mais volátil, por isso, quando uma empresa anuncia a mudança de CEO, os mercados vão reagir de uma forma mais drástica, resultando num preço por ação ainda mais baixo, aumentando, ainda mais, o rendimento dos dividendos. Ainda durante este período, as empresas têm mais incentivos para reter os seus ganhos, por isso, os CEOs decidem pagar menos dividendos. Os resultados também referem que a mudança de CEO tem um efeito positivo nos dividendos por ação e no rendimento dos dividendos depois da crise financeira.
Assim, na prática este trabalho evidencia, pela primeira vez, que a mudança de CEO tem um impacto significativo na política de dividendos das empresas, contribuindo assim para a literatura já existente sobre mudança de CEO e política de dividendos.
Classificação JEL: G32; G35; G40
Palavras-chave: Mudança de CEO, Política de Dividendos, Rendimento dos Dividendos,
iii
Acknowledgements
I would like to refer that this dissertation would not be possible without the help and support of certain people.
Firstly, I would like to thank my family for all the support and guidance throughout this process. I am very grateful for all they have done for me, the principles that they gave me helped me to surpass so many obstacles during my personal and academic life.
Then I wanted to thank all my friends, for all their support, friendship and kindness. I would like to thank my supervisor, Professor Victor Barros, for all the guidance, patience and ideas that led to this final work. I am very grateful that I had the chance to work and learn with Professor Victor.
I would also like to thank Professor Joaquim Sarmento for his help throughout the making of this dissertation. His clear insights and comments were a huge contribution to this study.
Finally, I gratefully acknowledge Nova School of Business and Economics (Nova SBE) for allowing me to access the databases that they have available, which had a crucial contribution to this dissertation and its success.
iv
List of Abbreviations
DPS – Dividend Per ShareDY – Dividend Yield
DPR – Dividend Payout Ratio
EBIT – Earnings Before Interest and Taxes EBT – Earnings Before Taxes
ROE – Return on Equity NPM – Net Profit Margin FE – Fixed Effects RE – Random Effects
v
List of Tables
Table 1 – Tests ... 26
Table 2 – Correlation Matrix ... 26
Table 3 – Dependent Variables Descriptive Statistics ... 26
Table 4 – Independent Variables Descriptive Statistics ... 26
Table 5 – Variables Description ... 27
Table 6 – Determinants of dividend payments ... 28
Table 7 – Determinants of the amount paid in dividends ... 29
Table 8 – Determinants of the amount paid in dividends by period ... 30
Table 9 – Determinants of DPS stability in the sample period and its sub-periods ... 31
Table 10 – Determinants of DY stability in the sample period and its sub-periods ... 32
Table 11 – Determinants of DPR in special cases ... 33
Table 12 – Determinants of dividends paid, DPS and DY lagged variables of first order ... 34
Table 13 – Determinants of dividends paid, DPS and DY lagged variables of second order ... 35
Table 14 – Influence of CEO_ in dividends paid, DPS and DY ... 36
Table 15 – Literature Review Summary Table of Theoretical Papers of CEO turnover ... 37
Table 16 – Literature Review Summary Table of Theoretical Papers of Dividend Policy ... 37
Table 17 – Literature Review Summary Table of Empirical Papers of CEO turnover ... 38
Table 18 – Literature Review Summary Table of Empirical Papers of Dividend Policy ... 46
List of Figures
Figure 1 – CEO turnover by year ... 23Figure 2 – CEO turnover by Business Sector... 23
Figure 3 – Dependent Variable analysis: DPS ... 24
Figure 4 – Residuals analysis: DPS ... 24
Figure 5 – Dependent Variable analysis: DY ... 25
vi
Index
ABSTRACT ... I RESUMO ... II LIST OF TABLES ... V LIST OF FIGURES ... V 1 INTRODUCTION ... 1 2 LITERATURE REVIEW ... 3 2.1 DIVIDEND POLICY ... 3 2.2 CEO TURNOVER ... 42.3 MAIN DETERMINANTS OF DIVIDEND POLICY ... 7
2.3.1 CEO Overconfidence ... 7
2.3.2 Control Variables ... 8
3 RESEARCH HYPOTHESES ... 10
4 EMPIRICAL ANALYSIS ... 11
4.1 DATA AND METHODOLOGY ... 11
4.2 VARIABLES ... 12
4.2.1 Dependent Variables: Dividend Policy ... 12
4.2.2 Descriptive statistics ... 12 4.3 REGRESSION MODELS ... 13 4.3.1 Principal Research ... 13 5 RESULTS ... 15 5.1 PRINCIPAL RESEARCH ... 15 5.2 ADDITIONAL RESEARCH ... 17 5.2.1 Methodology ... 17 5.2.2 Results ... 17
6 CONCLUSIONS AND FUTURE RESEARCH ... 19
7 REFERENCES ... 21
1
1 Introduction
CEO turnover and dividend policy have been studied by academics throughout the years. However, to the best of our knowledge, the relationship between these two topics have never been studied. Therefore, in this empirical research it will be studied what the influence of CEO turnover in firms’ dividend policy is.
In past studies, the relationship between CEO turnover and corporate performance has been a predominant topic. Empirical evidence found that support for a negative relation between firms’ performance and CEO turnover (Puffer & Weintrop, 1991; Kang & Shivdasani, 1995; Huson et al., 2004). Such effect is even more negative if performance is measured relative to firms’ peers (Kang & Shivdasani, 1995; Jenter & Kanaan, 2015).
Despite the large focus on performance, other factors can also influence the frequency of a CEO being dismissed from its role. According to Brickley (2003), the CEO’s age can also be an important factor in explaining CEO turnover. Murphy and Zimmerman (1993) stated that the probability of a CEO turnover happening is higher when the CEO’s age is 64 or 65. Such results are expected as such is considered to be a “normal retirement age”. Murphy (1999) results also suggested that CEOs were most likely to be dismissed at such ages. Consequently, this variable allows researchers to distinguish whether a CEO turnover is forced or occurs due to retirement reasons. Likewise, the composition of the Board of Directors can also influence CEO turnover. As independent or outside directors are added to firms’ boards, the independence towards the CEOs increases. Consequently, the probability of a CEO being replaced increases (Hermalin & Weisbach, 1998; Brickley, 2003).
Dividends are a portion of the firm’s earnings that is distributed to its shareholders. Nonetheless, firms’ management can decide to retain such earnings. Miller and Modigliani (1961) argued that in perfect capital markets conditions, dividend policy does not affect firms’ valuation. However, current markets are not perfect, containing market imperfections that can affect a firm’s dividend policy. Consequently, the existence of asymmetric information, a market imperfection, suggests that firms may pay dividends to mislead, and attract new investors. Additional studies indicate that paying dividends may reduce conflicts of interest between stakeholders, and thus, reduce agency costs within a firm structure.
Finally, a predominant market imperfection in today’s world is the existence of taxation. Dividends are seen as a return for shareholders and usually face higher tax rates than capital gains. However, shareholders still prefer to receive dividends as a form of compensation.
2
Thus, in this dissertation, we intend to clarify whether CEO turnover influences dividend policy. This work will contribute in understanding an unexplored topic and provide a clearer insight into how CEO turnover and dividend policy are related.
In this research, we considered companies that were listed on the S&P 500 Index over the period 2004-2017. The data related with CEO turnover was subject to manual adjustments using CEOs’ biographies and news which will be explained in detail in further sections. Several variables were selected to represent different factors that can influence both CEO turnover and dividend policy. A detailed description of these variables will be shown in further sections.
The empirical evidence suggests that CEO turnover increases firms’ dividend yield by about 0.2%. Moreover, CEO turnover that occurs during 2008 and 2012 has a positive effect on the dividend yield of 0.5%. Also, during the same period (2008-2012) firms have more incentives to retain their earnings. Therefore, CEOs decide to pay fewer dividends. Evidence also indicates that CEO turnover has a positive effect on dividend per share and dividend yield after the financial crisis.
Thus, this work contributes to practice since evidences, for the first time, that CEO turnover has a significant impact on firms’ dividend policy, contributing to the existing literature of both CEO turnover and dividend policy. This dissertation is organized as follows: in section 2 the most relevant literature regarding dividend policy and CEO turnover is reviewed. Section 3 contains the research hypotheses. The data, methodology, and regression models considered in our empirical analysis are shown in section 4. The results obtained from the econometric models are displayed and discussed in section 5. Finally, section 6 contains the main conclusions and limitations of our work, as well as proposed future researches.
3
2 Literature Review
2.1
Dividend Policy
The theory of dividend policy is grounded in the research of Lintner (1956) and Miller and Modigliani (1961). Miller and Modigliani (1961) argued that in perfect capital markets conditions, i.e., markets with the absence of taxes, transaction costs or other market imperfections, dividend policy does not affect a firm’s value. However, since the current market is not perfect in practice, several market imperfections can influence firms’ dividend policy.
The market imperfections displayed in this dissertation are mainly taxation, agency costs and asymmetric information. In this research, taxation has a predominant effect since dividends and share repurchases face different tax implications, and, therefore, investors will face different tax rates whether they receive dividends or repurchase shares. Black (1976) focused on why corporations pay dividends. Investors see dividends as a return for risking their wealth in a firm. Therefore, corporations continue to pay dividends to reward their shareholders and to encourage others to invest in their firm. However, dividends and share repurchases have different tax implications as investors face different tax rates. By repurchasing shares, shareholders will be taxed at the capital gain tax rate, whereas, if the corporation pays cash dividends, shareholders will be taxed according to the dividend tax rate. Usually, cash dividends are taxed at a higher rate than capital gains. Thus, logically, shareholders would prefer to repurchase shares instead of receiving dividends, if they act rationally. An additional tax advantage in share repurchases is that taxes on capital gains can be deferred until shares are sold, while taxes on dividends must be paid immediately. However, besides all the tax’s disadvantages for investors, corporations continue to pay dividends. This is called the dividend puzzle (Black, 1976).
Another market imperfection that influences dividend policy concerns agency costs. Jensen (1986) pointed out that conflicts of interest between managers and shareholders take place when the firm obtains an excess free cash flow. Managers can then use the excess free cash flows to pay dividends, rather than waste it in projects that may yield a low return. Jensen (1986) also documented that agency costs can be reduced by using debt. The use of debt leads to a decrease in the cash flow available for spending at the discretion of managers, therefore, minimizing agency costs within a firm. Thus, the author concludes that debt and dividends can be seen as substitutes. Fewer dividends are paid out to shareholders, as more debt is used.
The existence of asymmetric information in current markets is another market imperfection. Asymmetry is considered a market imperfection since firms can decide to pay out
4
dividends in order to mislead investors. According to the Dividend Signalling Theory, when a company announces that will increase in the level of dividend, investors take that as an indicator of positive future growth opportunities. Therefore, since managers’ information about their firm and its prospects is far more superior than the information obtained by outside investors, they can mislead investors by paying out dividends in order to attract new capital.
2.2
CEO turnover
Over the past years, the Board of Directors and CEO turnover has been the subject of extensive conceptualisation and empirical research. For most entities, having a Board of Directors is a legal requirement that must satisfy several regulations. A Board of Directors can be considered a financial institution that helps to solve an agency problem within most organisations. CEOs tend to satisfy their interests (compensation and other benefits), and the Board of Directors has the role of ensuring that shareholders’ interests are satisfied (Puffer & Weintrop, 1991). However, in order to achieve their organisational goals, an understanding of the role of the other is crucial to succeed. The Board of Directors must maintain its independence to monitor the CEO’s decisions effectively, and whether decide to replace or to keep him as CEO.
A model was developed related to the Board of Directors that takes into consideration both CEO and board perspectives (Hermalin & Weisbach, 1998). Hermalin and Weisbach, (1998) documented that a board’s independence depends on the CEO’s bargaining power inside the board. When CEOs perform well, the board’s independence declines. Otherwise, outside members are added to the Board of Directors, increasing the board’s independence, which implies a higher probability of a CEO being dismissed. Therefore, board composition can influence the frequency of CEO turnover.
An additional study was made regarding the performance-turnover relation, in which evidence was found that performance is more relevant in the early stages of a CEO tenure (Dikolli et al., 2014). Furthermore, the tenure of departing CEOs influences firm operational performance since if a departing CEO’s tenure is too short, the firm may not have completely recovered from the previous turnover (Shen & Cannella, 2002). Evidence was also found that deteriorating firm performance leads to a management turnover (Huson et al., 2004), as the results show that financial performance tends to decrease before top management turnovers.
The uncertainty behind a CEO’s quality decision making creates a demand for considerable performances in order to diminish such uncertainty. Thus, the higher the CEO tenure, the lower is the level of independence of the board (Dikolli et al., 2014). Such evidence supports Hermalin and Weisbach (1998) model results.
5
Based on the model created by Hermalin and Weisbach (1998), the following results were found: CEOs who perform poorly have a higher likelihood of being replaced; CEO turnover is more sensitive to corporate performance when the board has a higher level of independence; outside independent members are added to the board after poor performances; board independence diminishes over a CEO tenure; accounting performances are better indicators of management turnover than stock prices performance; a CEO that is fired on private information basis should cause an adverse reaction in stock-prices. Alternatively, a CEO that is fired based on public information should have a positive impact in stock-prices; CEO’s compensation should be insensitive to previous bad performances but sensitive when CEOs perform exceptionally well.
As previously mentioned, a higher number of outside directors leads to higher independence of the Board of Directors, which causes an increase in the probability of CEO turnover. Empirical research also concluded that the sensitivity of CEO turnover to corporate performance increases with more outsiders in the boardroom (Brickley, 2003).
An additional study was made in which it was examined if corporate performance, measured as the difference between actual performance and boards’ performance expectations, was a better indicator of CEO turnover. The results showed a negative relation between corporate performance and CEO turnover. Using performance measures that reflect the board’s expectations, the relation is even more negative (Puffer & Weintrop, 1991). Also, Puffer and Weintrop (1991) documented that there is a turnover when annual reported earnings per share fall short of Board of Directors’ expectations.
Regarding executive compensation, analyses have shown that payment levels are higher and are less sensitive to corporate performance in larger corporations (Murphy, 1999). Moreover, evidence was found that upcoming CEOs take a “big bath” since market-adjusted account accruals are lower in the fiscal year in which the CEO is replaced by his or her successor (Murphy & Zimmerman, 1993). Additionally, research and development (R&D), advertising, capital expenditures expenses and accruals are lower during actual CEO turnover than in years -5 to -2 and years +1 to +5. CEOs have incentives to decrease these expenses in their last years in the role to increase reported accounting earnings and inflate their compensation (Murphy & Zimmerman, 1993).
CEOs also have other incentives to boost compensation. They can distort their firms reported financial statements by using accounting procedures to create statements that reflect an overly positive view of the firm’s current financial position. These mechanisms of earnings management take advantage of how accounting rules are established and creates financial reports with inflated earnings. Hence, earnings management is strongly associated with subsequently
6
forced CEO turnovers, but it is not significantly associated to voluntary turnovers (Hazarika et al., 2012).
Such results indicate that earnings management is negatively related to CEO tenure (Hazarika et al., 2012). Earnings management leads to a higher likelihood that the CEO will be fired over the short-run. Under the same statistical testing, similar results were found for CFOs (Hazarika et al., 2012).
There are additional factors that can explain CEO turnovers. Evidence was found that low stock returns in the firm’s industry and market increase the likelihood of a forced CEO turnover (Jenter & Kanaan, 2015). Additionally, it was found that peers’ performance influences the turnover of CEOs that are underperforming their competitors (Jenter & Kanaan, 2015). The same results were documented for Japanese firms, in which top executive turnover is negatively related to corporate performance and more negatively associated if performance is measured relative to its peers (Kang & Shivdasani, 1995). Also, nonroutine turnovers are significantly associated with industry-adjusted return on assets, excess returns and negative pre-tax earnings (Kang & Shivdasani, 1995). Such authors found improvements in firms’ performances after a forced turnover, but no evidence was found relative to routines turnovers. Another factor that can influence the probability of CEO turnover is CEO overconfidence. CEOs with low and high levels of optimism face a greater risk of being dismissed than moderately optimistic CEOs do (Campbell et al., 2011). Low-optimism CEOs have a 50% to 112% higher probability of facing a forced turnover than moderately optimistic CEOs have, whereas, high-optimism CEOs have a 28% to 99% greater probability of a forced turnover compared with moderately optimistic CEOs.
Furthermore, other variables seem relevant to explain CEO turnovers. Evidence was found that the age of the CEO is statistically significant in explaining CEO turnovers (Murphy, 1999). The research was conducted from 1970 until 1995 and concluded that CEOs were most likely to leave their corporations at ages 64/65. Such results are expected since ages 64/65 can be considered “normal retirement ages”. Moreover, evidence was found that underperforming executives tend to leave at younger ages: 34% of CEOs who were underperforming left the company before age 60. It was also documented that CEOs had a higher likelihood of being replaced by outside hires rather than internal promotions. The results suggest that the likelihood of CEO turnover is higher when the CEO’s age increases, and when it reaches its normal retirement age (Murphy & Zimmerman, 1993).
In most recent years, a study was made regarding CEO turnover behaviour from 1992 to 2007 for a sample of large US firms (Kaplan & Minton, 2012). Between 1992 and 1999, CEO turnover increased to 12.6% resulting in an average CEO tenure of, approximately, 8 years.
7
Whereas, since 2000, CEO turnover has increased to 16.8%, implying an average tenure of almost 6 years (Kaplan & Minton, 2012).
Another stream of literature examined internal and external monitoring mechanisms from the 1970s until the mid-1990s (Huson et al., 2001). Evidence was found that forced CEO turnover and the outside succession have increased throughout the sample period. As expected from the previous study (Kaplan & Minton, 2012). Also, these results show that the sensitivity of forced turnovers to corporate performance did not change over the period in question.
A more recent study was made regarding CEO turnover and performance sensitivity in both private and public firms. Evidence shows that public corporations tend to have higher CEO turnover and more performance sensitivity to the turnover compared with private firms (Gao et al., 2017). The authors concluded that the reason for such difference was investor myopia. In public firms, investors tend to focus more on short-term outcomes instead of being long-term oriented. They prefer short-term earnings rather than an increase in firms’ long-term value. Thus, the results provide an insight that public firms fire CEOs earlier than the optimal, implying that public CEOs suffer a more demanding short-term performance.
A CEO turnover can also signal changes in future corporate decisions. Non-voluntary management changes, initiated by the Board of Directors, and normal retirement at age 65 lead to divestitures of poorly-performing assets (Weisbach, 1995). Likewise, an investment made that does not fit the current firm’s assets could lead to CEO turnover.
Once a CEO turnover happens, there are 3 types of successors: contenders (executive that has the support and approval of the board); followers (successors who follow a CEO’s ordinary retirement to continue and follow the existing strategies); outsiders. A negative association between outsider successor and post-succession operational performance (measured by ROA) was found (Shen & Cannella, 2002). The results indicate that senior executive turnover is positively associated with firms’ ROA after a contender succession but negatively related following outsider successions. The results show that CEO succession does not influence companies’ market performance in the long-run.
2.3
Main determinants of Dividend Policy
2.3.1 CEO Overconfidence
The relation between CEO and dividend policy is a topic that has not been extensively studied. Nevertheless, Deshmukh et al. (2013) develop a model of interaction between CEO overconfidence and dividend policy. The results indicate that an overconfidence CEO views
8
external funding as costly and, therefore, prefers to increase the firm’s financial slack for future investment needs by lowering the current dividend payout (Deshmukh et al., 2013). Thus, the level of dividend payout is lower in firms managed by overconfident CEOs. However, the reduction in dividends related to CEO confidence is higher in firms with lower growth opportunities and lower cash flows. In addition, a positive market reaction to a dividend-increase announcement is higher for firms with greater uncertainty about overconfidence.
2.3.2 Control Variables
Lintner (1956) was the pioneer of the modern understanding of dividend policy. He developed a theoretical model of corporate dividend behaviour, in which it is observed that the target payout ratio is a variable which affects payout decisions. He documented that dividends distributed are a result of net income and dividend payout policy. Also, firms are averse to reducing their dividend payout ratio even when they face an environment of scarcity (Lintner, 1956).
The model was later adjusted to examine the determinants of dividend payments by individual firms (Fama & Babiak, 1968). The results suggest that dividends and some measure of current profits are relevant variables in explaining dividend changes. It is documented that net income seems to be a better measure of profits than either cash flow or net income and depreciation as separate variables in the model.
Later on, Rozeff (1982) focused on growth, beta and agency costs as determinants of dividend payout ratios. He found that dividend payout is a significantly negative function of the firms’ past and expected future growth rate of sales. Also, evidence suggests that DPR is a significantly negative function of its beta coefficient and percentage of stock held by insiders. While the dividend payout ratio is positively associated with the firms’ number of common stockholders. Additionally, the future predicted growth variable is more important than past realized growth (Rozeff, 1982). The results suggest that companies with higher investment have lower dividend payout ratios as they use their excess earnings in growth opportunities rather than distribute it to shareholders as dividends.
Thereafter, additional research was conducted regarding the determinants of corporate dividend policy (Alli et al., 1993). Results suggest that dividend payout ratios are negatively related with the cost of external funds (equity and debt), expected investment outlays and growth. Such results support Rozeff (1982)’s findings. It is also documented that a firm with a significant portion of their shares held by institutional investors has a higher payout ratio. Additionally, it was found that ownership dispersion does not affect corporate dividends. The argument that
9
dividends can be used to mitigate agency problems is also supported (Alli et al., 1993). Instead, companies with greater capital structure flexibility, i.e., that have easier access to capital markets are able to pay higher dividends (Alli et al., 1993). Although firms usually prefer to increase their financial slack rather than paying dividends.
Another market imperfection that influences dividend policy is investor protection. When investor protection is high, cash needs is more relevant in explaining dividend payout. Whereas, when there is poor investor protection, liquidity appears to be more critical than cash needs (Boţoc & Pirtea, 2014). The results also indicate that firm size and corporate governance are associated with higher dividend payouts. Liquidity and debt ratio have a positive effect on dividend payout, while cash need has a negative effect (Boţoc & Pirtea, 2014).
Furthermore, short-term investment horizons are negatively related with future propensity to pay dividends (Chang et al., 2016). The propensity to pay dividends increases with an increase in firm size, fixed assets ratio, firm age or profitability and decreases with an increase in leverage, cash ratio, sales growth or firm risk. While, the dividend payout ratio is positively related to firm size and cash ratio, and negatively associated with leverage, firm risk and profitability. These results support Botoç and Pirtea (2014) findings mentioned above.
In earlier empirical research, Ahmad et al. (2018) examine the determinants of dividend policy in Euronext 100 firms. Ahmad et. al (2018) measured dividend policy as the dividend yield. The results indicate that dividend yield is not related with firms’ profitability (contrary to expectations). Therefore, dividend yield may be higher for less-profitable firms. The results also suggest that firms’ growth is significant and negatively associated with dividend yield, and dividend yields are lower for larger firms (negative relation between dividend yield and firms’ size) (Ahmad et al., 2018). Ahmad et. al (2018) argue that leverage has a negative effect on the dividend yield of firms with a stable dividend payout ratio over time, while leverage may positively affect the dividend yield of firms with stable dividend per share.
Later studies were developed focusing on asymmetric information theories. The existing literature indicates that investors believe that when a company reports positive results and pays a substantial amount of dividends, this is a sign of an increase in future earnings (Dewenter & Warther, 1998). The results suggest that stock prices of Japanese firms react less strongly to dividend omissions and initiation announcements compared with US stock prices. Furthermore, Japanese firms tend to cut dividends in response to poor corporate performance more quickly than US firms.
10
Dividends and share repurchases are taxed at different tax rates (Lintner, 1956), and the relevant market imperfection to take into consideration in this dissertation is taxation. Chkir and Samir (2008) examined the relationship between taxation and corporate dividend policy, using two tax events that occurred in Canada. The first event accounted for the capital gains exemption that was reduced in 1987 from $500,000 to $100,000 and eliminated in 1994 (second event). The reduction of capital gains exemption was timid to boost the average dividend payout. However, the elimination of the capital gains exemption in 1994 had a considerably higher effect in increasing the level of dividend payouts. Chkir and Samir (2008) argued that taxation has an impact on corporate dividend policy and the changes in the tax reform proves the existence of a dividend clientele.
In Appendices are presented Table 15, Table 16, Table 17 and Table 18 with a more detail analysis regarding the papers used in the literature review and the main conclusions documented. Table 15 and Table 16 depict theoretical studies, while Table 17 and Table 18 show empirical researches used in this dissertation.
3
Research hypotheses
As the objective is to test the impact of CEO turnover in corporate dividend policy, it is crucial to select the independent variables to use. The selection of explanatory variables was based on previous literature and on data availability. There is not, to our knowledge, literature directly relating CEO turnover and firms’ dividend policy. However, the test hypotheses considered in this study are as follows:
A CEO turnover will most likely lead to changes in firms’ stock prices that will impact their dividend policy. Thus, we expected a positive association between CEO turnover and DY.
H1: CEO turnover is positively related with DY
Additionally, we expect a negative association between CEO turnover and dividend payments, DPS and DPR due to the use of “big bath accounting”. According to Murphy and Zimmerman (1993), future CEOs tend to boost future earnings by writing off unwanted operations and unprofitable divisions. Consequently, the earnings will drop, leading to a lower payout ratio.
H2: CEO turnover is negatively associated with dividend payments, DPS and DPR.
Also, during a financial crisis, usually, firms have more incentives to maintain resources rather than paying dividends to surpass a more difficult financial period, thus, leading to lower values of DPS and DPR. Additionally, during financial crisis periods stock prices tend to fall, which leads to higher dividend yields.
11
H3: Financial crisis is negatively associated with dividends paid by firms, DPS and DPR, and positively related with DY.
The literature is not consensual regarding the effect of leverage on dividend policy. According to Boţoc and Pirtea (2014), the debt ratio is positively related with DPR, since as more debt is used, more profit a firm gets and, consequently, more dividends can be paid. Whereas Chang et al. (2016) documented that leverage and dividend policy are negatively related. According to Jensen (1986), debt and dividends can be seen as substitutes by managers.
In this study, we decided to follow Chang et al. (2016) findings since their research focuses on US firms, whereas, Boţoc and Pirtea (2014) focus on emerging countries.
H4: Leverage is negatively related with firms’ dividend policy.
According to Chkir and Samir (2008), taxation has an impact on corporate dividend policy. However, the existing literature is not consensual regarding the exact effect of a tax rate in the dividend policy. Therefore, the effective tax rate can be positively or negatively associated with dividend policy.
H5: Tax Rate is positively or negatively related with firms’ dividend payments, DPS, DPR and DY.
4
Empirical Analysis
4.1
Data and Methodology
The study focuses on companies listed on the S&P 500 Index. The initial sample was composed of 505 companies, with a sample period from 1992 to 2018. The data was extracted using financial datasets within Wharton Research Data Services (WRDS). For CEO turnover, it was used Compustat – Capital IQ, whereas, financial data was extracted from CRSP/Compustat Merged. Financial ratios were collected using Financial Ratios Suite by WRDS. After the data management, the sample used in this dissertation contains 394 firms between 2004 and 2017, equating up to 4,155 firm-year observations. Adjustments were made to CEO turnover data by using CEOs’ biographies present on Bloomberg and news available online1.
1 A CEO turnover was not considered in companies that had co-CEOs, and one co-CEO was dismissed from its role
in the following period. CEO turnover was also disregarded from the observations for firms with one CEO, co-CEOs in the next period and, thereafter, a unique CEO, with a common CEO in the periods in question.
12
4.2
Variables
4.2.1 Dependent Variables: Dividend Policy
The main goal of this dissertation is to answer the question: “Does CEO turnover influence dividend policy?”, therefore, various dependent variables were chosen in order to represent firms’ dividend policy fully.
Firstly, we aim to test the relationship between CEO turnover and a firm paying or not dividends, as in other studies (Chang et al., 2016). Consequently, we set the dependent variable as a dummy that equals one if firms pay dividends and zero otherwise.
Secondly, we aim to estimate how CEO turnover is related to the amount of cash dividends paid by each firm. The dependent variables in the second specification are dividend yield (DY), dividend per share (DPS) and dividend payout ratio (DPR). These variables were already extensively used in empirical papers with similar goals, namely, Rozeff (1982), Alli et al. (1993), Chkir and Samir (2008), Boţoc and Pirtea (2014), and Ahmad et al. (2018).
Furthermore, we aim to analyse the level of dividends paid by corporations following CEO turnover. Therefore, additional dummy variables were created to represent the level of dividends, such as DPS Regular that equals one if the DPS is equal to the DPS from the previous period, and zero otherwise. DY Regular and DPR Regular were also created using the same method as DPS Regular.
Finally, as CEO turnover can have a delayed impact on corporate dividend policy, an additional analysis was carried out using lagged variables for the main dependent variables.
Table 5 shows a more detailed description of the dependent and independent variables used in this dissertation.
4.2.2 Descriptive statistics
The sample used for this dissertation contains about 487 CEO turnovers that occurred in the 394 firms included in our sample. The variable CEO was defined as a dummy variable, which takes the value one if there was a CEO turnover during the actual year of turnover and zero otherwise.
Additionally, Figure 1 shows how CEO turnover has behaved throughout the sample’s period. The vertical lines represent the financial crisis period, defined as previously mentioned (2008-2012). As seen in Figure 1, CEO turnover has been increasing during the last years and reached its all-time high in 2017.
13
In our study CEO takes the value one in the year of the turnover. However, in additional analyses the variable takes the value one for the years after the turnover (T+1 or T+2) to account for the fact that CEOs may not decide the future dividend policy immediately after the turnover. Also, in robustness checks, the variable CEO takes the value one from the year of the turnover onwards.
Before moving to the next section, we look at some descriptive statistics for our dataset, available on Tables 3 and 4. Table 4 presents the descriptive statistics for the independent variables. The high values of the Market-to-Book ratio correspond to firms that have a low reported book value of equity and net income, indicating that they are facing financial difficulties and near bankruptcy. Table 3 shows the descriptive analysis of dependent variables.
4.3
Regression Models
4.3.1 Principal Research
The data used in this dissertation is unbalanced panel data, in which we have to decide whether to use a fixed-effects model (FE) or a random-effects model (RE). Hence, the Hausman test was performed. For the models which have DPS and DY as dependent variables, the fixed effects model is recommended. Although, in order to have a completed comparison between all measures of dividend policy, FE and RE estimators will be used for each model, and the respective results will be presented.
Afterwards, to assess the joint significance of independent variables in our models, the Wald test was performed. The test concludes that the chosen explanatory variables are significant in explaining the behaviour of the dependent variables.
Additionally, the Ramsey and Specification link tests were performed to confirm if there were no omitted variables, and a p-value of zero was obtained. Therefore, our models appear to be biased by omitted variables. Results that we were expecting since we did not control for either corporate governance nor the CEO’s age. According to Murphy and Zimmerman (1993), Brickley (2003) and Huson et al. (2004) CEO’s age can be considered a significant variable in explaining CEO turnovers since would allow us to distinguish between forced turnovers and if the respective CEO left due to retirement reasons. We encourage further researches to include this variable as a control variable, although due to data availability, such was not considered. Also, due to lack of data, variables that control for corporate governance (e.g., institutional ownership) were not included in our models.
After that, in order to have unbiased results, we also tested for the presence of heteroskedasticity in our models. The Breusch and Pagan test was performed, to understand
14
whether the variance of the errors differs across observations [Var (εi|xi)]. In order to have
better-estimated results, we adjusted for heteroskedasticity by using robust standard errors.
Table 1 in the Appendices shows all test results performed during this dissertation. As seen in Table 2 in the Appendices, there is no multicollinearity in our models since there is no high degree of correlation between the independent variables used. VIF tests also support that there is no multicollinearity in the models used.
For those variables that are defined as a dummy, i.e., that equal one if a determined condition is true and zero otherwise, a Probit and Logit model will be used. A Probit and Logit model allows us to accurately measure the impact of the independent variables on a dummy variable. The major difference between both models lies in the assumption on the distribution of the errors. In the Logit model, the errors are assumed to follow the standard logistic distribution, whereas, for the Probit model, it is assumed that the errors follow a Normal distribution. In general, both models tend to lead to similar results, but both will be presented in order to have more accurate conclusions. Initially, we want to test the impact of CEO turnover on a firm paying or not dividends, thus, the following model was created:
D_dividendsit = β0 + β1CEOit + β2ln_Assetsit + β3ROEit + β4FinCrisisit + β5NPMit + β6Levit + β7MBRit + β8TaxRateit + εit
(1)
Where D_dividendsit is a dummy variable defined as one if a firm pays dividends and
zero otherwise.
An additional model was created using the regression (1), by creating an interaction between CEO and FinCrisis instead of controlling these variables separately. The variable referred is CEO_crisis,
We used the same approach when testing the impact of CEO in Dividends, which may be captured by either DPS or DY:
Dividendsit = β0 + β1CEOit + β2ln_Assetsit + β3ROEit + β4FinCrisisit + β5NPMit + β6Levit + β7MBRit + β8TaxRateit + εit
(2)
In order to test if CEO turnover influences a firm’s dividend policy before, during and after the financial crisis period, we ran three different regression models for each measure of corporate dividend policy (DPS and DY). To test the impact of a CEO turnover before the financial crisis, we ran the model for years before 2008. To test the effect during the financial crisis, we set the years between 2008 and 2012. Finally, to test the impact of CEO after the crisis, we ran the model for years after 2012.
15
As mentioned in section 4.2.1, we want to estimate the impact of CEO turnover in dividends’ stability level. Thus, Dividends_regular dummy variable was defined, which represents either DPS_regular or DY_regular.
Dividends_regularit = β0 + β1CEOit + β2ln_Assetsit + β3ROEit + β4FinCrisisit + β5NPMit + β6Levit + β7MBRit + β8TaxRateit + εit
(3)
Moreover, CEO turnovers can have a delayed impact in a firm’s dividend policy, since upcoming CEOs just know the company and its current financial situation and, therefore, dividends can only shift in the following periods. We test this hypothesis by creating two final regressions for each dependent variable, replacing D_dividends, Dividends (DPS and DY) with lagged versions of first and second order.
Finally, in order to fully test whether CEO turnover influences firms’ dividend policy from the year of the actual turnover onwards, the variable CEO_ was created and defined as a dummy variable that equals one from the year that the turnover occurred onwards and zero otherwise.
5 Results
5.1
Principal Research
In this section, by observing the estimation results obtained through our regression models, the question “Does CEO turnover influence dividend policy?” will be answered. Tables 6-14 show the results that will support such conclusions.
Firstly, Table 6 and Table 7 present the results for the regressions models (1) and (2), respectively. The results show that CEO turnover is statistically significant in explaining the behaviour of firms’ DY (Table 7). More precisely, CEO turnover will increase corporate DY by about 0.2%. This may be justified by market stock prices adjustment once a CEO turnover takes place, i.e., stock prices may decrease after a CEO turnover, leading to an increase in dividend yields. However, a CEO being replaced does not have a statistical influence on the dividends paid by firms and DPS (Table 6 and 7, respectively). Table 6 also displays that the natural logarithm of total assets (proxy for company size) is positively related with dividend payments, which supports the findings from previous studies. Mature firms may not find good projects to invest that add value, therefore, they decide to pay dividends instead. Leverage is also positively associated with firms paying dividends which means that companies pay dividends using cash from leverage events, contrary to our expectations (Table 6).
16
The period of the financial crisis is statistically significant at all significance levels and seems to have a negative effect on D_dividends and DPS (Table 6 and 7, respectively), and a positive effect on firms’ DY (Table 7). As mentioned in section 3, these results are expected since corporations tend to have more incentives to retain resources rather than paying to shareholders during this period, and their stock prices tend to suffer a negative shift during a stressful financial period. Leverage is also statistically significant and has a positive impact in DPS and DY behaviour (Table 7), supporting the trade-off theory of capital structure (the more debt is used, the more profit the firm generates, and more dividends can be paid), contrary to our expectations. Additionally, as stated in the previous section, it was created an interaction between CEO turnover and the financial crisis period. The purpose is to test the exact influence of a CEO turnover during the financial crisis, in a period where, usually, corporations face a more demanding financial period. Table 7 shows that CEO turnovers occurring during 2008 and 2012 are statistically significant and increase the DY by 0.5%, while, decreasing the dividends paid by corporations (Table 6). By comparing both results, it is concluded that CEO turnovers more than doubled its effect on dividend yields, during the financial crisis. Even though the dividends paid by corporations decrease, CEO turnover leads to an increase in DY. This is expected since, during a financial crisis, stock prices are considerably more volatile and, therefore, CEO turnovers lead to even lower stock prices.
Afterwards, it was tested how CEO turnover influences dividend policy across different periods in our sample. We decided to test it in three events: before the financial crisis took place, during the financial crisis and after it. Table 8 suggests that CEO turnover is statistically significant and has a positive effect on DPS and DY after 2012.
Since we already study the influence of CEO turnover in DPS and DY, we decided to cover the impact of CEO turnover in the stability of DPS and DY. The reason why we created two additional dummy variables, DPS_regular and DY_regular, was previously explained in section 4.2.1. Tables 9 and 10 display the estimated results for the regression model (3). We conclude that CEO turnover has no influence on dividend per share and dividend yield stability levels, in all the sample time frame and its sub-periods.
Additionally, we also study the hypothesis of whether there is a delayed effect of CEO turnover in firms’ dividend policy. A CEO turnover could not have an immediate effect and take longer to have an impact on how dividend policy is managed. We test this hypothesis by replacing the dependent variables with lagged versions of it, of first and second order, respectively. The estimation results suggest that there is no delay influence of CEO turnover on how dividends are managed (Table 12 and 13). As mentioned previously, we also test the influence of CEO turnover
17
in firms’ dividend policy from the year of the actual turnover onwards and the results show that CEO turnover has no effect on how firms managed their dividend policy in the years following the turnover (Table 14).
5.2
Additional Research
5.2.1 Methodology
We decided to present DPR as an additional analysis section since there are some statistical limitations regarding its respective regression model.
The same method was used as in the previous analysis. Firstly, it was performed a Hausman test, and it was concluded that a RE estimator would provide more accurate estimates. However, both FE and RE models will be shown as in the previous section.
Thereafter, the Wald test was computed, suggesting that the independent variables have some limitations in explaining firms’ DPR behaviour. This is the reason why we decide to include DPR as an additional analysis while reinforcing that the results will likely be biased.
The Ramsey and Specification link tests were also performed in this section. For DPR statistical evidence was also found that supports the existence of heteroskedasticity. Therefore, robust standard errors will be used to adjust for heteroskedasticity.
The dividend payout ratio is the proportion of earnings paid out as dividends to shareholders. Therefore, for this additional analysis, we decided to use a different approach. We wanted to test how CEO turnover influences firms’ DPR in special cases. Consequently, variables were created to represent when a firm’s DPR is negative, between zero and one, and higher than one.
5.2.2 Results
As additional research, we decided to test if CEO turnover influences DPR in special cases, i.e., when DPR is negative, between zero and one, and higher than one. However, we reinforce that this particular study has some statistical limitations.
Table 11 shows that CEO turnover is statistically significant and has a positive effect when DPR is negative and higher than one. Such results can be explained by the existence of dividend clientele. Even though firms generate negative earnings, they keep paying out dividends to avoid changing their dividend policy due to the existence of dividend clientele. Consequently, when DPR is negative, CEO turnover leads to an even lower DPR.
Likewise, a DPR higher than one suffers an increase when a CEO turnover takes place. A DPR higher than one means that the firm is paying out more to its shareholders than the earnings coming in. These corporations want to maintain their DPS regular due to the existence of dividend
18
clientele and, therefore, avoid changing how they manage their dividend policy. The existence of dividend clientele alongside the focus on restructuring the firm, leads to paying more as dividends than the earnings received. Moreover, when there is a stable DPR (between zero and one), CEO turnover has negative effects on firms’ DPR. However, bear in mind that these are biased estimation results and, thus, these conclusions may have some limitations.
19
6 Conclusions and Future Research
The purpose of our research is to identify whether CEO turnover influences firms’ dividend policy. We focused on firms listed on the S&P 500 Index with a sample period between 2004 and 2017 and with up to 4,155 firm-year observations. About 487 turnovers occurred in the 394 firms included in our sample.
Based on past studies regarding CEO turnover and dividend policy, several variables were chosen as explanatory variables. The remaining explanatory variables served to control for some firms’ characteristics like size, growth opportunities, corporate profitability and the effective tax rate that the respective firms faced.
In this dissertation, we intend to clearly understand how CEO turnover influences firms’ dividend policy. To do so, we used Logit and Probit models when testing for a dummy variable and a FE and RE models otherwise.
Evidence was found that CEO turnover is statistically significant and increases firms’ DY by about 0.2%. Such shift can be explained by market price adjustments when a CEO turnover occurs, i.e., stock prices will decrease once a CEO turnover occurs, leading to an increase in dividend yields. No evidence was found that supports the existence of an association between CEO turnover and either the dividends paid by firms and their DPS.
We, additionally, confirmed that the financial crisis is significant and has a negative effect on DPS and in dividend payments, while, a positive effect on DY behaviour which is consistent to our expectations. We expected these results since during this period companies tend to have more incentives to retain its earnings; therefore, a negative impact on the dividends. Also, during financial stressful periods, firms stock prices tend to be lower and more volatile, increasing the dividend yields.
Evidence was also found that CEO turnovers that occurred between 2008 and 2012 have an even higher positive effect on DY and lead to a decrease in the dividends paid. As explained above, during the financial crisis stock prices are more volatile, therefore, when a firm announces a CEO turnover, the market will react less smoothly and lead to even lower stock prices, increasing, even more, the dividend yields. Also, it is also expected that between 2008 and 2012, CEOs decide to pay fewer dividends.
Thereafter, as explained in the previous section, we decided to display how CEO turnover influences dividend policy across sub-periods in our sample. Three events were defined: before
20
2008, during the financial crisis and after 2012. The estimation results suggest CEO turnover is statistically significant and has a positive effect on DPS and DY after the financial crisis.
We then decided to identify the effect of CEO turnover in DPS and DY stability, i.e., in DPS regular and DY regular. We concluded that CEO turnover is not statistically significant in explaining DPS and DY stability. Likewise, the results show that there are no delay effects of CEO turnover in dividend policy, i.e., CEO turnover does not take longer to impact how firms manage their dividend policy.
Finally, since there are some limitations regarding DPR evidence, we decided to include it as additional research in this dissertation. We took a different approach and test how CEO turnover influences DPR in three special cases: negative DPR, between zero and one, and higher than one. The results display that CEO turnover is statistically significant and has a positive effect when DPR is negative due to the existence of dividend clientele as corporations avoid changing their dividend policy even if they generate a net loss. Furthermore, when there is a more stable DPR (between zero and one), CEO turnover has a negative effect on firms’ DPR. Finally, when the DPR is higher than one, a CEO turnover will lead to an increase in the dividend payout ratio, i.e., firms will pay out more to its shareholders than the earnings obtained. These results are also explained by the existence of dividend clientele and the focus on restructuring the firm, leading to paying out more as dividends than the earnings generated. Nevertheless, we reinforce that since these are most likely biased results, such conclusions may have some limitations.
Thus, this work contributes to practice since evidences, for the first time, that CEO turnover has a significant impact on firms’ dividend policy, contributing to the existing literature of both CEO turnover and dividend policy.
We suggest that future researches on this particular topic use the CEO’s age and variables to represent firms’ corporate governance as control variables. According to previous studies, the CEO’s age is a significant variable in explaining CEO turnovers. And, such variable would allow distinguishing turnovers between forced turnovers and turnovers that occur during a CEO “normal retirement age”. We invite future research to include both variables; however, due to data availability, such were not included in our research. Also, we encourage future research to analyse in detail the impact of financial crisis in both CEO turnover and firms’ dividend policy. Additionally, an alternative to our approach would be to use a negative binomial regression, as most dependent variables have values near zero. It would be interesting to investigate this further and compare the results between the two approaches.
21
7 References
Ahmad, N. G., Barros, V., & Sarmento, J. M. (2018). The determinants of dividend policy in Euronext 100. Corporate Ownership & Control, 15(4), 8–17.
Alli, K. L., Khan, A. Q., & Ramirez, G. G. (1993). Determinants of Corporate Dividend Policy: A Factorial Analysis. The Financial Review, 28(4), 523–547.
Black, F. (1976). The Dividend Puzzle. The Journal of Portfolio Management, 2(2), 5–8. Boţoc, C., & Pirtea, M. (2014). Dividend Payout-Policy Drivers: Evidence from Emerging
Countries. Emerging Markets Finance and Trade, 50(sup4), 95–112.
Brickley, J. A. (2003). Empirical research on CEO turnover and firm-performance: a discussion. Journal of Accounting and Economics, 36(1–3), 227–233.
Campbell, T. C., Gallmeyer, M., Johnson, S. A., Rutherford, J., & Stanley, B. W. (2011). CEO optimism and forced turnover. Journal of Financial Economics, 101(3), 695–712. Chang, K., Kang, E., & Li, Y. (2016). Effect of institutional ownership on dividends: An
agency-theory-based analysis. Journal of Business Research, 69(7), 2551–2559.
Chkir, I., & Samir, S. (2008). Taxation & dividend policy: new empirical evidence. Corporate Ownership & Control, 5(4-4), 432-439.
Deshmukh, S., Goel, A. M., & Howe, K. M. (2013). CEO overconfidence and dividend policy. Journal of Financial Intermediation, 22(3), 440–463.
Dewenter, K. L., & Warther, V. A. (1998). Dividends, Asymmetric Information, and Agency Conflicts: Evidence from a Comparison of the Dividend Policies of Japanese and U.S. Firms. The Journal of Finance, 53(3), 879–904.
Dikolli, S. S., Mayew, W. J., & Nanda, D. (2014). CEO tenure and the performance-turnover relation. Review of Accounting Studies, 19(1), 281–327.
Fama, E. F., & Babiak, H. (1968). Dividend Policy: An Empirical Analysis. Journal of the American Statistical Association, 63(324), 1132–1161.
Gao, H., Harford, J., & Li, K. (2017). CEO Turnover–Performance Sensitivity in Private Firms. Journal of Financial and Quantitative Analysis, 52(02), 583–611.
Hazarika, S., Karpoff, J. M., & Nahata, R. (2012). Internal corporate governance, CEO turnover, and earnings management. Journal of Financial Economics, 104(1), 44–69.
Hermalin, B. E., & Weisbach, M. S. (1998). Endogenously Chosen Boards of Directors and Their Monitoring of the CEO. The American Economic Review, 88(1), 96–118.
Huson, M. R., Malatesta, P. H., & Parrino, R. (2004). Managerial succession and firm performance. Journal of Financial Economics, 74(2), 237–275.
22
Huson, M. R., Parrino, R., & Starks, L. T. (2001). Internal Monitoring Mechanisms and CEO Turnover: A Long-Term Perspective. The Journal of Finance, 56(6), 2265–2297. Jensen, Michael C, 1986. Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers.
The American Economic Review 76, 323-329.
Jenter, D., & Kanaan, F. (2015). CEO Turnover and Relative Performance Evaluation: CEO Turnover and Relative Performance Evaluation. The Journal of Finance, 70(5), 2155– 2184.
Kang, J.-K., & Shivdasani, A. (1995). Firm performance, corporate governance, and top executive turnover in Japan. Journal of Financial Economics, 38(1), 29–58.
Kaplan, S. N., & Minton, B. A. (2012). How Has CEO Turnover Changed?: How Has CEO Turnover Changed? International Review of Finance, 12(1), 57–87.
Lintner, J. (1956). Distribution of Incomes of Corporations Among Dividends, Retained Earnings, and Taxes. 18.
Miller, M. H., & Modigliani, F. (1961). Dividend Policy, Growth, and the Valuation of Shares. The Journal of Business, 34(4), 411.
Murphy, K. J. (1999). Chapter 38 Executive compensation. Em Handbook of Labor Economics (Vol. 3, pp. 2485–2563).
Murphy, K. J., & Zimmerman, J. L. (1993). Financial performance surrounding CEO turnover. Journal of Accounting and Economics, 16(1–3), 273–315.
Puffer, S. M., & Weintrop, J. B. (1991). Corporate Performance and CEO Turnover: The Role of Performance Expectations. Administrative Science Quarterly, 36(1), 1.
Rozeff, M. S. (1982). Growth, Beta and Agency Costs as Determinants of Dividend Payout Ratios. Journal of Financial Research, 5(3), 249–259.
Shen, W., & Cannella, A. A. (2002). Revisiting the Performance Consequences of CEO Succession: The Impacts of Successor Type, Postsuccession Senior Executive Turnover, and Departing CEO Tenure. Academy of Management Journal, 45(4), 717–733.
Weisbach, M. S. (1995). CEO turnover and the firm’s investment decisions. Journal of Financial Economics, 37(2), 159–188.
23
8 Appendices
Figure 1
–CEO turnover by year
Source: Author
Figure 2
–CEO turnover by Business Sector
24
Figure 3
–Dependent Variable analysis: DPS
Figure 3 presents the standardised normal probability plot and histogram of dependent variable Dividend Per Share (DPS).
Source: Author
The behaviour of the standardised normal probability plot is explained by a considerable amount of observations with lower values of DPS.
Figure 4
–Residuals analysis: DPS
Figure 4 shows the Kernel density estimation on the squared residuals as the normal density and the residuals’ standardised normal probability plot.
25
Figure 5
–Dependent Variable analysis: DY
Figure 5 presents the standardised normal probability plot and histogram of dependent variable Dividend Yield (DY).
Source: Author
Figure 6
–Residuals analysis: DY
Figure 6 shows the Kernel density estimation on the squared residuals as the normal density, and the residuals’ standardised normal probability plot.
26
Table 1 – Tests
Test DPS DY DPR Wald 0.000 0.000 0.286 Hausman 0.000 0.003 0.330 Ramsey 0.000 0.000 0.115 Specification 0.000 0.000 0.084 Heteroskedasticity 0.000 0.000 0.000 In the table above it is shown the p-value for each test with DPS, DY and DPR as dependent variables.Table 2
–Correlation Matrix
CEO CEO_ ln_Assets ROE FinCrisis NPM Lev MBR TaxRate CEO 1.000 CEO_ 0.640* 1.000 ln_Assets 0.012 -0.037* 1.000 ROE -0.010 0.004 -0.024 1.000 FinCrisis -0.036* -0.021 -0.044* -0.050* 1.000 NPM -0.033* -0.005 0.118* 0.468* -0.062* 1.000 Lev 0.004 0.008 0.059* 0.003 -0.050* -0.082* 1.000 MBR -0.002 -0.010 -0.241* 0.369* -0.131* -0.089* 0.107* 1.000 TaxRate 0.013 0.001 0.018 0.007 -0.001 0.005 0.010 -0.005 1.000 * p < 0.1
Table 3
–Dependent Variables Descriptive Statistics
Variable Obs Mean Std. 1st Q Median 3rd Q Min Max
D_dividends 4,508 0.824 0.381 1.000 1.000 1.000 0.000 1.000 DPS 4,508 0.981 1.091 0.210 0.770 1.460 0.000 27.030 DY 4,508 0.019 0.029 0.006 0.016 0.027 0.000 1.428 DPR 4,507 0.317 2.075 0.047 0.260 0.468 -49.000 89.000 DPS_regular 4,508 0.256 0.436 0.000 0.000 1.000 0.000 1.000 DY_regular 4,508 0.149 0.356 0.000 0.000 0.000 0.000 1.000 DPR_regular 4,508 0.149 0.356 0.000 0.000 0.000 0.000 1.000 DPR_negative 4,508 0.046 0.209 0.000 0.000 0.000 0.000 1.000 DPR_0and1 4,508 0.908 0.289 1.000 1.000 1.000 0.000 1.000 DPR_more1 4,508 0.046 0.210 0.000 0.000 0.000 0.000 1.000
Table 4
–Independent Variables Descriptive Statistics
Variable Obs Mean Std. 1st Q Median 3rd Q Min Max
CEO 4,508 0.108 0.310 0.000 0.000 0.000 0.000 1.000 CEO_ 4,508 0.228 0.420 0.000 0.000 0.000 0.000 1.000 ln_Assets 4,508 9.799 1.449 8.730 9.701 10.668 4.407 14.761 ROE 4,508 0.152 0.184 0.083 0.140 0.208 -2.198 2.345 FinCrisis 4,508 0.360 0.480 0.000 0.000 1.000 0.000 1.000 NPM 4,508 0.077 0.204 0.048 0.085 0.133 -4.038 0.573 Lev 4,505 0.241 0.150 0.132 0.231 0.339 0.000 0.960 MBR 4,508 3.429 4.270 1.460 2.484 4.000 0.175 100.000 TaxRate 4,157 0.316 2.156 0.220 0.321 0.392 -70.167 56.413
